Sun Aug 30 00:12:17 CEST 2015

Functions of signals

What does it actually mean?

(s->m(s,a)) -> (s->m(s,b)) -> (s->m(s,c))

Give two state machines, a new state machine can be constructed.  The
latter machine operates by executing a step in the two input machines,
performing some internal computation, and producing an output step and
state update.  The state update is (necessarily) composed of the input
states and possibly some internal state private to the internal

The main question remains: why are signal processors not Kleisli
arrows (from instantaneous values to signals)?

There seems to be someting interesting in creating a linear
composition law, where kleisli composition would be linear convolution
of impulse responses.  This seems dual to the local, comonadic view
(composing time-delayed filters).

Does it make sense to make compositions of "non-linear" impulse